Cryptographic research has taken a significant step forward with a new paper examining the internal mechanics of the Keccak sponge function—the foundational primitive behind the SHA-3 cryptographic hash standard. Authored by Danilo Gligoroski of the Norwegian University of Science and Technology (NTNU), the paper titled "Beyond the Event Horizon of Keccak, Extending the Observation from 2011 and applications" builds upon insights originally documented over a decade ago. The study introduces three novel theoretical theorems regarding sponge capacities, provides verified full-round examples, and outlines a series of unconventional cryptographic protocols that leverage tunable sponge capacities.
The research revisits a foundational observation made in 2011, which noted that a single path in the Keccak sponge—starting from a state whose capacity has all zero bits and returning to such a state—yields an explicit, infinite family of second preimages for any message spanning at least one full rate block. Finding this specific path carries a computational cost of approximately two raised to the power of $c$ divided by two ($2^c/2$). For a standard implementation like SHA-3-256, which utilizes a capacity of $c = 512$ bits, this cost translates to $2^256$ operations. In cryptographic terms, this threshold effectively lies "beyond the event horizon," rendering it computationally infeasible for standard security parameters. However, Gligoroski’s new work expands this observation into generalized theorems and explores practical applications when the capacity parameter is deliberately adjusted.
Expanding the 2011 Framework Through Three Novel Theorems
The newly published paper advances the 2011 observation by introducing three comprehensive theorems that broaden how capacity states and internal loops function within the Keccak sponge construction.
The first theorem establishes that the capacity does not need to be restricted to all-zero bits; it can possess any arbitrary value $x$. By constructing a two-block meet-in-the-middle loop that moves from $x$ back to $x$—a process achieved at a heuristic cost of roughly two times $2^c/2$—researchers can generate the exact same unlimited family of second preimages without requiring any additional search operations. Connectors leading to and from the value $x$ can be established at approximately the same computational cost.
Building upon this foundation, Theorem 2 extends the concept further to encompass multi-block $k$-block loops as well as complete cycles of the capacity map. The research demonstrates that a single-state bridge—defined as a state $S$ where the capacity part of the forward transformation equals the capacity part of the inverse transformation, denoted as $textcap(f(S)) = textcap(f^-1(S))$—can similarly generate an infinite family of second preimages, though this approach incurs a cost of roughly $2^c$.
Theorem 3 shifts focus toward collision generation, introducing a method for achieving cheap collisions and a bridge-amplified multicollision framework. According to the paper, a single inner collision combined with a single connector can yield equal-length Joux-style multicollisions of size $2^k$ for any arbitrary value of $k$. This occurs in a single step, bypassing the traditional requirement for $k$ separate collision searches as outlined in Antoine Joux’s classic multicollision attacks. To the best of the author’s knowledge, this represents the first such construction for generic sponge functions where the search cost remains entirely independent of $k$.
To validate these theoretical breakthroughs, the author provides fully verified, full-round examples executed on both Keccak-f[200] and Keccak-f[1600] instances configured with a reduced capacity of $c = 32$ bits. Furthermore, the algebraic results stated throughout the research have been machine-checked using Lean 4, an interactive theorem prover, ensuring mathematical rigor. The author has also made reproducible experiments and open-source proof-of-concept code publicly available to the cryptographic community.
Novel Protocols and Unkeyed Applications
Moving beyond theoretical cryptanalysis, the second part of Gligoroski’s paper introduces a series of novel cryptographic protocols. These designs are built upon the deliberate asymmetry between the computationally intensive process of minting Keccak bridges and the relatively straightforward task of forging or verifying them.
Among the proposed applications is a proof-of-work mechanism featuring a hardness-adaptive capacity, allowing systems to dynamically adjust security parameters based on computational demand. The paper also outlines dual-capacity commitments anchored securely to the real SHA-3-256 digest, treating bridges themselves as zero-knowledge-committed tokens. Another notable proposal is a self-enforcing bounty market designed to test and evaluate reduced-round security margins.
Crucially, these protocols represent some of the first unkeyed, Keccak-based constructions that intentionally manipulate the sponge capacity so that the generic bridge cost becomes a feasible and tunable hardness parameter, rather than an insurmountable barrier. While these applications offer intriguing new architectural pathways for protocol design, the author explicitly notes that they remain theoretical proposals and that their comprehensive, end-to-end security properties remain an open question for peer review and further cryptographic analysis.
The complete research paper, along with its associated BibTeX citation data and open-source materials, is publicly accessible via the Cryptology ePrint Archive under Paper 2026/2671.






